Post

CT Dataset Collection: A Comprehensive Reference

A curated reference of 90+ publicly available CT datasets for medical and industrial imaging research, including download links, dataset sizes, and usage notes.

Posted Updated
CT medical imaging dataset collection
CT medical imaging dataset collection
By YuXuan Yan
11 min read
CT Dataset Collection: A Comprehensive Reference

Last Updated: February 2026 Total Datasets: 90+ publicly available datasets Coverage: Medical CT imaging, industrial CT non-destructive testing, surface defect detection

All datasets are for academic research purposes only. Always verify the license agreement and privacy policy before use. Commercial use requires explicit authorization.

📋 Table of Contents

  1. Medical CT Datasets
  2. Industrial CT and Defect Detection Datasets
  3. Comprehensive Resource Platforms
  4. Important Notes

🏥 Medical CT Datasets

Whole-Body Multi-Organ Segmentation

1. TotalSegmentator 系列

  • Name: TotalSegmentator / TotalSegmentator v2
  • Scale: 1204例 CT图像 (v1) / 1228例 (v2)
  • Annotations: 104个解剖结构 (v1) / 117个结构 (v2)
  • Split: 1082训练 + 57验证 + 65测试
  • Notes: 目前最大的三维医学图像分割数据集
  • Download: https://zenodo.org/records/10047292
  • GitHub: https://github.com/wasserth/TotalSegmentator
  • Paper: Wasserthal et al., Radiology: Artificial Intelligence, 2023

2. AbdomenCT-1K

  • Scale: 1000+ CT扫描
  • Notes: 腹部多器官分割，多中心数据
  • Annotations: 详细的器官级注释
  • Download: https://github.com/JunMa11/AbdomenCT-1K
  • Paper: https://arxiv.org/abs/2203.02739 (Zenodo: Part 1: https://zenodo.org/record/5903099 | Part 2: https://zenodo.org/record/5903846 | Part 3: https://zenodo.org/record/5903769)

Lung Datasets

3. LIDC-IDRI (肺图像数据库联盟)

  • Scale: 1018例 CT扫描
  • Notes: 诊断和肺癌筛查胸部CT，标注病灶
  • Annotations: 4位经验丰富的放射科医生标注
  • Applications: 肺癌检测和诊断CAD方法
  • Download: https://www.cancerimagingarchive.net/collection/lidc-idri/
  • Organized by: 美国国家癌症研究所(NCI)

4. MSD肺癌分割

5. LoLa11

  • Task: 肺叶分割
  • Applications: 放射治疗规划
  • Download: https://zenodo.org/records/4708800

6. StructSeg2019

  • Scale: 50例 CT
  • Task: 肺癌放射治疗危及器官分割
  • Annotations: 左右肺、脊髓、心脏、食道、气管
  • Download: https://structseg2019.grand-challenge.org/
  • Applications: 放射治疗规划

7. QIN Lung CT

  • Applications: 肺部定量成像分析
  • Multi-center: 多家医疗机构数据
  • Download: https://www.cancerimagingarchive.net/collection/qin-lung-ct/

8. COVID-19 CT相关

  • COVID-19-CT SCAN IMAGES
  • BIMCV COVID-19: 201例 CT扫描，包含注释
  • CT Images in COVID-19: 2482例 COVID/非COVID二分类
  • Download:
    • https://tianchi.aliyun.com/dataset/93666
    • https://github.com/BIMCV-CSUSP/BIMCV-COVID-19
    • https://www.cancerimagingarchive.net/collection/ct-images-in-covid-19/

9. 肺炎相关数据集

  • IEEE 8023 COVID-19: 20组扫描，肺部和感染区标注
    • 下载: https://github.com/IEEE8023/covid-chestxray-dataset
  • RSNA肺炎检测: Kaggle竞赛，26684张训练数据
    • 下载: https://www.kaggle.com/c/rsna-pneumonia-detection-challenge/data

Liver Datasets

10. Sliver07

  • 用途: 肝脏分割基准
  • Description: 常与LiTS合并使用
  • Download: https://zenodo.org/records/2597908

11. 3D-IRCADB

  • Scale: 20例患者(10女10男)
  • Notes: 75%肝肿瘤患者的3D CT
  • Contains: 肝脏大小、肿瘤位置等详细信息
  • Applications: 肝脏分割软件测试
  • Download: https://www.ircad.fr/research/data-sets/liver-segmentation-3d-ircadb-01/

12. CHAOS (腹部器官)

  • Modality: MR、CT
  • Task: 腹部器官分割
  • Download: https://zenodo.org/records/3431873
  • Scale: 120例 (80训练+40测试)

13. MSD肝脏血管分割

  • Task: 肝脏和肝血管精细分割
  • Applications: 肝脏手术规划
  • Download: http://medicaldecathlon.com/
  • Source: Medical Segmentation Decathlon任务05

14. TCGA-LIHC

  • Applications: 肝细胞癌研究
  • Download: https://www.cancerimagingarchive.net/collection/tcga-lihc/

Brain Datasets

15. BraTS系列

  • Name: BraTS 2013/2015/2021等
  • Task: 脑肿瘤分割和生存分析
  • Modality: MR (主要)
  • Scale: 400+ 例患者
  • Challenge: MICCAI年度挑战
  • Download: https://www.med.upenn.edu/cbica/brats/

16. Kaggle RSNA-MICCAI Brain Tumor

  • Download: https://www.kaggle.com/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/data

17. MSD脑肿瘤分割

  • Modality: MR
  • Task: 脑胶质瘤分割
  • Download: https://drive.google.com/drive/folders/1HqEgzS8BV2c7xYNrZdEAnrHk7osJJ–2
  • Source: Medical Segmentation Decathlon任务01

18. MSD海马体分割

  • 结构: 脑海马体细分割
  • Download: https://drive.google.com/drive/folders/1HqEgzS8BV2c7xYNrZdEAnrHk7osJJ–2
  • Source: Medical Segmentation Decathlon任务09

Abdominal Organ Datasets

20. FLARE 2022 (Fast Abdominal Lesion Recognition)

  • Scale: 2300例 CT (50例标注+2000例无标注+50验证+200测试)
  • Source: 20+ 医学中心
  • Task: 13种腹部器官分割
  • Notes: 跨中心、多供应商、多模态、多期、多疾病
  • Challenge: MICCAI 2022
  • Website: https://flare22.grand-challenge.org/Dataset/

21. AMOS 2022 (Abdominal Multi-Organ Segmentation)

  • Scale: 500例 CT + 100例 MR
  • Source: 多中心、多供应商、多模态
  • Annotations: 15种腹部器官
  • Download: https://zenodo.org/records/7262581

22. WORD (Whole abdominal Organs Recognition in CT)

  • Scale: 150张 CT
  • Split: 100训练 + 20验证 + 30测试
  • Annotations: 16种腹部器官详细标注
  • Website: https://github.com/HiLab-git/WORD

23. RAOS dataset

  • Website: https://github.com/Luoxd1996/RAOS
  • Note: RAOS real CT dataset unzip password: “RAOS@2023”; synthetic MRI: “raos@2023”. MICCAI2024 accepted; cite WORD (MedIA2023) when using.

24. Pancreas-CT

  • Scale: 80例 (53男27女)
  • Notes: 对比增强3D CT，胰腺手动标注
  • Download: https://www.cancerimagingarchive.net/collection/pancreas-ct/

25. Kidney-Tumor (KiTS)

  • Task: 肾脏和肾肿瘤分割
  • Scale: 300+ 例
  • Download: https://kits19.grand-challenge.org/
  • GitHub: https://github.com/neheller/kits19

Chest CT Datasets

27. CT-RATE (Chest CT Reports And Text)

  • Scale: 50,188例 (47,149训练+3,039验证)
  • Notes: 第一个包含CT图像、诊断报告、异常标签的大规模数据集
  • Patients: 21,304例不同患者
  • Annotations: 18种异常条件
  • Time span: 2015年5月 - 2023年1月
  • Download: https://huggingface.co/datasets/ibrahimhamamci/CT-RATE

28. RAD-ChestCT

  • Scale: 3,630例 (初始发布，完整数据35,747例)
  • Patients: 19,661例成人
  • Source: Duke医学中心
  • Download: https://zenodo.org/records/6406114

29. Chest CT-Scan Images

  • Scale: 1000例
  • 分类: 肺癌 - 4类 (正常、腺癌、大细胞癌、鳞状细胞癌)
  • Download: https://tianchi.aliyun.com/dataset/93929

30. CheXpert

  • Scale: 224K图像 + 36M文本token
  • Annotations: 肺部、胸部、心脏疾病
  • Download: Stanford AIMI

31. MIMIC-CXR

  • Scale: 377,110张胸部X光 (227,835例研究)
  • Notes: DICOM格式，带自由文本放射报告；满足HIPAA Safe Harbor要求
  • Download: https://physionet.org/content/mimic-cxr/2.1.0/
  • Note: 需要request，需要导师的reference

Other Organ Segmentation

32. SegRap 2023 (头颈部)

  • Scale: 200例 CT
  • Annotations: 45类头颈部危险器官 + 2类鼻咽癌和淋巴结
  • Applications: 头颈癌放射治疗规划
  • Website: https://segrap2023.grand-challenge.org/
  • Challenge: MICCAI 2023

33. HaN-Seg (Head and Neck)

  • Scale: 42例 CT & MRI
  • Annotations: 30类头颈部危险器官
  • Download: https://zenodo.org/records/7442914

34. ToothFairy 2023 (牙科)

  • Scale: 443例 CBCT (153例密集+290例稀疏标注)
  • Task: 下颌神经体素级分割
  • Download: https://ditto.ing.unimore.it/toothfairy/

35. Colon-CT & 肠道相关

  • CT Colonography: TCIA 结肠癌CT数据
    • 下载: https://www.cancerimagingarchive.net/collection/ct-colonography/

37. 椎间盘与脊椎相关

  • xVertSeg: 骨折椎体分割和分类 — https://lit.fe.uni-lj.si/en/research/resources/xVertSeg/
  • VerSe: 脊椎分割和标记 — https://github.com/anjany/verse

🏭 Industrial CT and Defect Detection Datasets

Non-Destructive Testing Datasets

38. 2DeteCT (2D可扩展实验CT)

  • Scale: 5000个2D CT切片
  • Notes: 范扇束CT，实验数据(非模拟)
  • Applications: 低剂量重建、有限角度采样、束硬化伪影减少、超分辨率、分割
  • Scan modes: 3种 (高保真、低剂量、束硬化)
  • Paper: Nature Scientific Data, 2023
  • Download: Slices1-1000: https://zenodo.org/records/8014758Slices1001-2000: https://zenodo.org/records/8014766Slices2001-3000: https://zenodo.org/records/8014787Slices3001-4000: https://zenodo.org/records/8014829Slices4001-5000: https://zenodo.org/records/8014874Slices(OOD): https://zenodo.org/records/8014907

39. LoDoPaB-CT

  • Scale: 40,000+ CT切片
  • Source: LIDC/IDRI数据库 (~800患者)
  • Notes: 用于CT重建的ML任务，包含投影和image
  • Download: https://zenodo.org/records/3384092
  • Paper: https://arxiv.org/abs/2106.06542
  • Online HDF5 viewer: https://hdfviewer.com/

40. 医学MNIST-3100

  • Scale: 58,954张医学图像
  • Modality: 脑部CT、手部CT、胸部CT、腹部CT、乳腺MRI、胸部X光
  • Applications: 医学图像分类基准
  • Download: https://www.kaggle.com/datasets/andrewmvd/medical-mnist

41. CBCTLiTS (合成CBCT肝脏)

  • Scale: 合成CBCT + 配对高质量CT
  • Task: 肝脏和肝肿瘤分割
  • Quality levels: 5种质量级别 (高视觉质量到严重伪影)
  • 下载: https://www.kaggle.com/datasets/maximiliantschuchnig/cbct-liver-and-liver-tumor-segmentation-train-data

🌐 Comprehensive Resource Platforms

Key GitHub Projects

ProjectURLCoverage
Awesome-Medical-Datasethttps://github.com/openmedlab/Awesome-Medical-Dataset100+ datasets, CT/MRI/X-ray/ultrasound/pathology
医学影像数据集集锦https://github.com/linhandev/dataset71+ datasets, Chinese descriptions
Awesome-Medical-Image-Segmentation-Datasethttps://github.com/ziyangwang007/Awesome-Medical-Image-Segmentation-Dataset200+ segmentation datasets
Medical-Imaging-Datasetshttps://github.com/m-aryayi/Medical-Imaging-DatasetsWith license and download info

Data Access Platforms

76. The Cancer Imaging Archive (TCIA)

  • URL: https://www.cancerimagingarchive.net/
  • Scale: 500+ 公开数据集集合
  • Data types: CT、MRI、PET、X光等

77. Grand Challenges

  • URL: https://grand-challenge.org/
  • Contains: 100+ 医学影像挑战赛

78. Kaggle医学影像数据集

  • URL: https://www.kaggle.com/
  • Scale: 数百个医学相关数据集

79. Zenodo学术数据库

  • URL: https://zenodo.org/
  • Notes: DOI引用和版本控制

80. OpenDataLab

  • URL: https://opendatalab.com/
  • Notes: 中文开源数据平台，国内访问速度快

83. Medical Segmentation Decathlon

  • URL: http://medicaldecathlon.com/
  • Scale: 10个不同分割任务
  • 器官: 脑、心脏、肝脏、肺、胰腺、胰腺肿瘤、肠膜肿瘤、脾脏、骶骨、结肠

📌 Important Notes

Data Usage Guidelines

All datasets are for academic research purposes only. Strictly comply with each dataset’s license agreement. Commercial use requires explicit authorization.

HIPAA Privacy Protection:

  • Medical data has been de-identified
  • Complies with HIPAA Safe Harbor requirements
  • Does not contain protected health information (PHI)

Citation Guidelines:

  • Always cite the original paper when using a dataset
  • Provide data source and institutional acknowledgment

Data Type Notes

CharacteristicMedical CTIndustrial CT
FormatDICOM or NIfTI (.nii.gz)PNG, JPG, TIFF, or proprietary
Resolution512×512 or higher200×200 to 4096×4096
Slice thicknessTypically 1–5 mmVaries by application
Annotations3D segmentation masksSegmentation masks or bounding boxes

🔧 Common Tools

CategoryTools
VisualizationITK-SNAP, 3D Slicer
Python processingSimpleITK, MONAI, pydicom, nibabel
Data augmentationimgaug, albumentations
Format conversiondcmread (DICOM), nibabel (NIfTI)

Dataset Selection Guide

By Task:

  • Segmentation: TotalSegmentator, AMOS, FLARE series
  • Classification: ChestCT-Scan, COVID-CT
  • Detection: LIDC-IDRI, DeepLesion, KiTS

By Scale:

  • Large-scale (>1000 scans): TotalSegmentator, FLARE, CT-RATE
  • Medium-scale (100–1000): LiTS, LIDC-IDRI, BraTS
  • Small-scale (<100): Specific organs or rare diseases

Special Requirements:

  • Multi-modal: AMOS (CT+MRI), CHAOS
  • Multi-center: FLARE, AMOS, CT-RATE
  • Industrial/NDT: 2DeteCT, LoDoPaB-CT

🔗 Quick Links

ResourceURL
Awesome-Medical-Datasethttps://github.com/openmedlab/Awesome-Medical-Dataset
TCIAhttps://www.cancerimagingarchive.net/
Grand Challengeshttps://grand-challenge.org/
OpenDataLabhttps://opendatalab.com/
TotalSegmentatorhttps://zenodo.org/record/6802614
LIDC-IDRIhttps://www.cancerimagingarchive.net/collection/lidc-idri/
CT-RATEhttps://huggingface.co/datasets/ibrahimhamamci/CT-RATE

Last updated: February 2026 — 90+ datasets indexed. Part of my CT reconstruction research notes series.

Disclaimer: This document is for academic reference only. All dataset copyrights belong to their respective institutions. Before using any dataset, confirm compliance with its license agreement and privacy policy.

Research, Datasets
ct-imaging open-data medical-imaging benchmark
This post is licensed under CC BY 4.0 by the author.